ORIGINAL RESEARCH

published: 26 July 2021

doi: 10.3389/fmats.2021.680895

Multi-Objective Optimization Methods

for Designing Low-Carbon Concrete
Mixtures
M.A. DeRousseau *, J.R. Kasprzyk and W.V. Srubar III
Department of Civil, Environmental, and Architectural Engineering, University of Colorado Boulder, Boulder, CO, United States

Concrete mixtures are complex material systems with a multitude of characteristics that
decision-makers may deem important. These characteristics can include economic,
environmental, mechanical, and durability-related properties of a concrete mixture.
However, traditional concrete mixture design typically employs long-standing
heuristics, which satisfy requirements for physical characteristics but are unable to
minimize specific characteristics, such as the cost or carbon footprint of the concrete
Edited by:
Brant Walkley,
mixture. This work considers these performance characteristics by implementing
The University of Sheffield, simulation-optimization as a new paradigm for designing concrete mixtures. The utility
United Kingdom
of the simulation-optimization framework is tested for several concrete design case studies
Reviewed by:
that simultaneously consider compressive strength, embodied carbon, service life, and
Marlon Augusto Longhi,
Federal University of Rio Grande do cost. Results from these scenarios demonstrate that the local conditions of the case study
Sul, Brazil dictate the most important parameters of the simulation-optimization (i.e., relative
Xinyuan Ke,
University of Bath, United Kingdom
constituent costs, in situ service-life conditions). Out of all other parameters,
Maria Criado, constituent cost and service-life conditions impact the set of optimal concrete mixture
Instituto de Ciencias de la
designs in terms of the types and quantities of mixture ingredients that are utilized. We
Construcción Eduardo Torroja
(IETCC), Spain present a simulation-optimization framework that is demonstrated herein to be a holistic
Jiri Klemes, design tool that allows designers to quantify and visualize tradeoffs between critical
Brno University of Technology,
Czechia
concrete performance metrics. Such a tool can be used to precision-tailor low-carbon
*Correspondence:
concrete mixtures to the exact preferences of the designer.
M.A. DeRousseau
Keywords: concrete, multi-objective optimization, embodied carbon, low-carbon concrete, modeling
[email protected]

Specialty section: INTRODUCTION

This article was submitted to
Structural Materials, Concrete is a complex, heterogeneous composite material, and a myriad of properties must be
a section of the journal considered in determining appropriate concrete mixture proportions. For instance, for a given
Frontiers in Materials
application, a concrete mixture must provide sufficient compressive strength, and oftentimes must
Received: 16 March 2021 resist degradation from multiple environmental factors. In addition, due to concerns about climate
Accepted: 23 June 2021
change, there is increasing pressure for the cement and concrete industry to produce concrete with
Published: 26 July 2021
lower upfront embodied carbon, namely the carbon dioxide emissions associated with manufacture
Citation: and transportation of concrete (“Making Concrete Change”; Rodgers, 2018). Note that the term
DeRousseau MA, Kasprzyk JR and
“embodied carbon” or “carbon emissions” refers to the carbon dioxide emissions associated with
Srubar WV (2021) Multi-Objective
Optimization Methods for Designing
material manufacture. Low-carbon materials, therefore, refer to materials with low carbon dioxide
Low-Carbon Concrete Mixtures. manufacturing emissions. Finally, project budgets and profitability concerns require finding the least
Front. Mater. 8:680895. expensive concrete that meets all design criteria. Finding a concrete mixture design that can satisfy all
doi: 10.3389/fmats.2021.680895 of these requirements is thus quite challenging.

Frontiers in Materials | www.frontiersin.org 1 July 2021 | Volume 8 | Article 680895

DeRousseau et al. Multi-Objective Optimization of Low-Carbon Concrete

The vast majority of current concrete mixture design models it uses. The use of more advanced property models for
methodologies, such as the American Concrete Institute’s predicting economic, environmental, service life, and mechanical
Standard Practice for Selecting Proportions for Normal, properties may lead to more accurate and trustworthy mix design
Heavyweight, and Mass Concrete (ACI 211.1–91), rely on solutions. The second category of studies in the literature uses
engineering heuristics in order to determine the quantities of multi-objective optimization for analyzing highly specialized
each concrete constituent (American Concrete Institute, 1991). concrete applications such as concrete that has very high
Such proportioning methods have the benefit of simplicity but strength (Baykasoğlu, 2009) or concrete that incorporates
have several disadvantages. First, these heuristics used do not uncommon agricultural (Madurwar et al., 2015) or industrial
capture the true complexity of the relationship between mixture waste products (Promentilla et al., 2017); however, it is difficult to
design and desired properties. For example, in the ACI method, extend such studies to more broad applications since the
one achieves the appropriate 28-days compressive strength with a simulation models they use are developed using specialized
heuristic in which the target compressive strength is mapped to datasets.
the necessary water-to-cement (w/c) ratio. Compressive strength The primary research gap our work addresses is that the
is indeed influenced by the w/c ratio; however, it is also strongly existing literature has not considered economic,
affected by other variables including the total cement content and environmental, mechanical, and durability-related properties
the type, quantity, and composition of any supplementary simultaneously using multi-objective optimization; this
cementitious materials (SCMs) (Toutanji et al., 2004; Khatri research gap is discussed extensively in (DeRousseau et al.,
et al., 1995). Other relationships are similarly simplified, such 2018). We seek to expand upon the existing research by
as the heuristics for slump and durability, meaning that true value incorporating these properties in a simulation-optimization (as
of these variables often differs from the target value. Another illustrated in Figure 1) that can be used for a variety of concrete
drawback of traditional concrete design methods is that it is not applications and types. In this framework, concrete properties
currently possible to truly optimize (i.e., minimize or maximize) (i.e., compressive strength, embodied carbon, cost, and service
properties of interest, while meeting design considerations related life) are simulated using modeled relationships (i.e., equations or
to the fresh-state, hardened-state, durability, economic, and predictive models) that link mixture proportions to material
environmental properties of concrete. For instance, traditional performance. Subsequently, optimal mixture designs are found
mixture design methods lack the ability to identify the mixture in an iterative fashion using a state-of-the-art multi-objective
design with the lowest cost or lowest embodied carbon, given a set evolutionary algorithm called Borg. We apply the simulation-
of performance requirements (e.g., required strength, slump, optimization framework to three location-based case studies in
sufficient resistance to chloride-induced corrosion). order to illustrate how input parameters, such as concrete
To address these challenges, this study proposes multi- constituent costs, transportation distances, and chloride
objective optimization as a method for designing concrete exposure conditions affect optimal mixture designs. Three
mixtures with multiple design considerations. Application of location-based scenarios are selected to represent very different
multi-objective optimization to concrete mixture design possible design cases in 1) Miami, FL; 2) San Francisco, CA; and
requires the use of property models which link a concrete 3) Anchorage, AK. The purpose of the case studies is to apply the
mixture design to the values of critical concrete properties, simulation-optimization framework to possible design situations
such as cost, compressive strength, expected service life, and and to illustrate that resultant optimal mixture designs are highly
embodied carbon dioxide emissions of the concrete mixture. dependent on input parameters related to individual constituent
Application of this technique to concrete mixture design costs, environmental exposure conditions, temperature, and
would enable designers 1) to better predict the properties of a constituent transportation distances.
given mixture design with these advanced property models and 2)
to simultaneously optimize multiple concrete properties of
interest. METHODS
The use of multi-objective optimization for proportioning
concrete mixtures is relatively underutilized in the literature. Problem Formulation
What literature exists can be generally grouped into two To conduct multi-objective optimization, one must define the
categories. The first category uses multi-objective optimization problem formulation, which is composed of decisions, objectives,
to address generalized concrete mixture design using two or three constraints, and modeled relationships for the problem. Decisions
property-related objectives. For instance, Young et al. selected are a set of variables (xi ) related to the concrete mixture design
cost and compressive strength as objectives and elected to that the decision-maker can vary. Note that the multi-objective
constrain the optimal solutions by their embodied carbon optimization method employed herein requires explicitly defined
values. However, a simplified model for embodied carbon was upper and lower bounds on the decisions, which limit the search
utilized in which the embodied carbon was approximated as a space. In this problem, the decisions relate to the type and
linear function of the quantity of cement; this linear function quantity of mixture ingredients. The decisions include: the
ignores the impact of other concrete constituents and life cycle total quantity of cementitious materials (i.e., cement and any
stages of concrete manufacturing (e.g., constituent transportation SCMS), the water-to-binder (w/b) ratio (where binder content
and use phase carbonation) (Young et al., 2019). The outcome of includes the quantity of both Portland cement and
optimizing a design problem is only as good as the property supplementary cementitious materials), the coarse aggregate to

Frontiers in Materials | www.frontiersin.org 2 July 2021 | Volume 8 | Article 680895

DeRousseau et al. Multi-Objective Optimization of Low-Carbon Concrete

TABLE 1 | Decision variables upper and lower limits.

Decision variable Lower limit Upper limit Units

Total cementitious materials 200 450 kg

Water-to-binder (w/b) ratio 0.20 0.75 kg/kg
Coarse-aggregate-to-total-aggregate ratio 55 75 kg/kg
SCM choice n/a (categorical variable) n/a (categorical variable) n/a
SCM replacement percentage 0 30% for fly ash, 50% for slag %

proposed in Case Studies, the dominant degradation mechanism is

TABLE 2 | Objective functions.
assumed to be chloride-induced corrosion. An additional constraint
Objective Minimize/maximize Type of model relates to water-reducing admixture dosage, which is required when
Compressive strength Maximize Machine learning
the w/b ratio is less than 0.40 so that the concrete remains workable
Embodied carbon Minimize LCA model (Domone, 1998).
Cost Minimize Linear combination Lastly, the modeled relationships are equations (i.e., models)
that relate the decision variables to quantifiable values of the
objectives and constraints. This analysis employs four modeled
total aggregate ratio (c/a), SCM choice, and SCM replacement relationships which have been developed in other research for
percentage. quantifying compressive strength, embodied carbon, cost, and
Each of the decision variables allow for the calculation of the chloride-induced corrosion. These modeled relationships are
quantities of each mixture ingredient (i.e., cement, fly ash, slag, further discussed in Modeled Relationships.
water, coarse aggregate, fine aggregate, water-reducer, air-
entraining admixture), and the absolute volume method is
used to quantify the mixture ingredient quantities from the Multi-Objective Optimization With
decision variables (American Concrete Institute, 1991). Evolutionary Algorithms
Table 1 provides the allowable range of values for the decision Assuming minimization of all objectives, multi-objective
variables, which are selected using engineering judgement and the optimization is formalized as the equations below (Coello
limits imposed to avoid issues associated with degradation or et al., 2007).
concrete setting. For instance, limits are placed on the w/b ratio to Minimize:
avoid segregation and bleeding (for high w/b values) and to
F(x) 
f1 (x), . . . , fk (x) (1)
ensure full cement hydration (for low w/b values). Similarly, to
avoid low values of early-age strength, limits are placed on the Subject to:
maximum percent replacement that is allowed for fly ash and slag,
gi (x)  0, i  {1, . . . , m} (2)
which are 30 and 50%, respectively, (Alabama Standard
Specifications, 2012). and:
Objectives (fi (x)) are the performance characteristics of a hj (x) ≥ 0, j  1, . . . , p (3)
mixture design that should be minimized (or maximized,
calculated via a functional relationship with the set of decision where each f(x) is an objective to be optimized and x is the
variables. In this analysis, we select three potentially competing n-dimensional decision space. gi (x) and hj (x) represent the
objectives (minimize cost, minimize embodied carbon, and inequality and equality constraints, respectively, and a
maximize compressive strength), which allows us to solution, x, is feasible if it meets all constraints. If there are
understand tradeoffs between the objectives. For instance, for conflicts between the objectives, a single solution cannot
each cost level, there will likely be tradeoffs between minimizing simultaneously optimize objectives. Thus, the goal of multi-
embodied carbon and maximizing compressive strength objective optimization is to generate a tradeoff set of solutions
objectives, since both compressive strength and embodied to the problem using the concept of Pareto-optimality. A vector of
carbon generally increases with total cement content. Table 2 decision variables, x, is Pareto-optimal if no other feasible vector
provides information on the objectives selected for this study; can minimize some objective without causing a simultaneous
justification for the inclusion of each objective function is increase in one or more other objectives. The result of multi-
provided in Modeled Relationships. objective optimization is, therefore, the values of decision
Constraints place limits on the values of the decisions or objectives variables from the Pareto optimal solutions, with their
of the problem. The difference between constraints and objectives is corresponding objective function values.
distinct; a constraint must be met for the design to be successful. In To search for the Pareto-optimal solutions, this analysis uses a
contrast, an objective is a design target where a lower value (or higher) multi-objective evolutionary algorithm (MOEA) called Borg,
is considered better. This analysis poses a constraint on concrete which was selected due to its auto-adaptive features and high
service life and requires that concrete must resist corrosion for a set performance on challenging multi-objective problems. (Reed
number of years as specified in Table 3). Note that for the case studies et al., 2013; Zatarain; Salazar and Jazmin, 2016). For each

Frontiers in Materials | www.frontiersin.org 3 July 2021 | Volume 8 | Article 680895

DeRousseau et al. Multi-Objective Optimization of Low-Carbon Concrete

TABLE 3 | Constraints.

Constraint Limit Type of model

Chloride-induced corrosion Must resist cracking for 50 or 150 years (depending on the scenario) Diffusion model
Water-reducing admixture If w/b ratio <0.40, use 0.703 ml superplasticizer per kg cementitious materials Heuristic

function evaluation, Borg stochastically generates and evaluates a needed to be tested to meet strength targets, and enables a
new solution (i.e., potential mixture design). A solution is non- reduction in factors of safety.
dominated if none of the objectives can be improved in value In this study, we utilize a validated machine learning model for
without degrading one or more of the other objective values when predicting the 28-days compressive strength, which was
compared to other existing solutions. During the search, the goal formulated by the authors and reported in (Mikaela
is to continue finding non-dominated solutions until a close DeRousseau et al.). In the study it was found that the random
approximation to the set of Pareto-optimal solutions is found. forest model performs best compared to all other models tested in
In this study, each scenario involved 50,000 function evaluations a machine learning pipeline in terms of root mean squared error,
and incorporated five random seeds. Using some number of mean absolute error, and coefficient of determination.
random seeds greater than one ensures that solutions were not an Consequently, this Python-based model is utilized in the
artifact of Borg’s initial parameterization. simulation-optimization framework as the objective function
for compressive strength. The inputs to the random forest
Modeled Relationships model are quantities of the mixture ingredients, which are
In the following sections, each of the modeled relationships for within the bounds of ingredient quantities found in the
calculating cost, compressive strength, embodied carbon, and training data. Since the decision variables are quantities such
chloride-induced corrosion are discussed in detail. Note that the as the w/b ratio, c/a ratio, percent replacement of SCMs, etc., we
cost, compressive strength, and embodied carbon models require use the absolute volume method to calculate the quantity of
the mass of the mixture ingredients as input variables. To mixture ingredients and subsequent compressive strength
calculate these values, we use the absolute volume method prediction (Wilson and Kosmatka, 2011). Choosing
(American Concrete Institute, 1991) applied to the decision compressive strength as an objective rather than a constraint
variables, since the decision variables are quantities such as the allows us to quantify tradeoffs made with the other objectives for
w/b ratio and SCM replacement percentage. each incremental increase in compressive strength.

Cost Model Embodied Carbon Model

To quantify the cost of a unit volume of concrete, the costs of The purpose of the upfront embodied carbon model is to quantify
each mixture ingredient are simply summed as shown in Eq. 4, the embodied carbon emissions for a functional unit of concrete
where Ctotal is the total cost of 1 m3 of concrete, Ci are the as a function of the mixture ingredient quantities. The embodied
costs per tonne of each mixture ingredient, and xi are the carbon model accounts for the life cycle modules of A1-A3 and
quantities of each mixture ingredient per unit of concrete B1 (i.e., emissions from raw material supply, transport,
(tonne/m3). This linear cost model is adapted by identifying manufacturing of products, and concrete carbonation). In this
unit costs of each individual mixture constituent in bulk in study, we utilize a model formulated and published by the authors
the location of each case study. The cost coefficients also in (MA DeRousseau et al., 2019) in which the embodied carbon
include cost of transportation to the ready-mix plant. was modeled as the sum of the material processing,
However, additional costs due to construction, maintenance, transportation, and manufacturing emissions, minus the sum
and end-of-life are not included here because it is assumed of the emissions due to carbonation during use and end-of-life for
that for a given application, these costs will be equivalent all ingredients for 1 m3 of concrete, as shown in Eq. 5.
for all alternatives. Note that the simulation-optimization n n
framework allows for mixture constituent costs to vary ECtot   (ci x)i + (xi di t) + m + s (5)
geographically, which is one reason why the set of Pareto- i i
optimal solutions varies for each case study.
In this equation, ci are the embodied carbon coefficients for each
Ctotal  Ci xi (4) constituent due to material processing (in kgCO2e/tonne), xi is
the quantity of each mixture ingredient, di is the transportation
distance for each mixture constituent, t is the transportation
Compressive Strength Model emissions factor (in kgCO2e/tonne/km) for the method of
The 28-days compressive strength of concrete is a critical design transportation, m is the embodied carbon associated with
parameter for reinforced concrete structures. Accurate numerical concrete manufacturing (in kgCO2e/m3), and s is the quantity
estimation of the 28-days compressive strength of concrete is of CO2 sequestered during the use and end-of-life stages of a
desirable because more precise prediction provides assurance of concrete in kgCO2e/m3 of concrete, which is a function of the
concrete quality, reduces the number of concrete batches that are quantity and CaO content of cementitious materials in the

Frontiers in Materials | www.frontiersin.org 4 July 2021 | Volume 8 | Article 680895

DeRousseau et al. Multi-Objective Optimization of Low-Carbon Concrete

TABLE 4 | Material design parameters (constant throughout case studies). kept constant among all three case studies, including the decision
Parameter Value Units variables, objectives, and constraints described in Problem
Formulation. In addition, we assume a set of input parameters
Concrete type Structural concrete — representing the constituent qualities (e.g., aggregate size, specific
Cover depth 70 mm
gravity) and the concrete design (e.g., rebar cover depth). These
Coarse aggregate specific gravity 2.50 kg/m3
Coarse aggregate transportation distance 27.5 km parameters are consistent for all case studies and provided in
Coarse aggregate cost 30 $/tonne Table 4.
Fine aggregate (sand) specific gravity 2.63 kg/m3 Other parameters, deemed the variable input parameters, are
Fine aggregate transportation distance 21.0 kg different for each case study, because they reflect three different
Fine aggregate cost 30 $/tonne
Cement transportation distance 105.0 km
location-based concrete design scenarios. The variable input
Cement specific gravity 3.15 kg/m3 parameters include: the relative costs of ingredients,
Water reducer cost 0.29 $/kg transportation distances, chloride exposure conditions, and
Slag specific gravity 0.38 kg/m3 monthly average temperatures. These parameters all vary
Fly ash specific gravity 0.27 kg/m3
based on project location, with the potential to impact the set
of optimal mixture designs. Table 5 provides a summary of the
variable input parameters that change between each of the case
concrete and therefore dependent on the SCM content. studies, which are based in Miami, FL; San Francisco, CA; and
Subsequently, this model is used for quantifying concrete Anchorage, AK. Cost parameters were estimated via expert
embodied carbon in the simulation-optimization framework interviews and transportation distances were estimated based
where minimizing embodied carbon is an objective. on the existence of local material production. Note that monthly
DeRousseau et al. (MA DeRousseau et al., 2020) analyze the average temperature data for Anchorage is reported in the table
variability in emissions using statistical distributions and the along with the predicted monthly average temperature for the
mean values of the life cycle inventory distributions in the 150th year of service life to simulate climate change effects, as
study are utilized, except for transportation distance. For discussed for Case 3b in the following section (Alaska Regional
transportation distances, we use values relevant for each case Climate Projections, 2009). Table 6 provides a qualitative
study location as inputs for calculating embodied carbon. summary of these parameters.
For the Miami case study (Case 1), the purpose is 1) to
Chloride-induced Corrosion Model investigate a concrete design scenario in which there is a high
Chloride-induced corrosion occurs when chloride ions from the chloride exposure condition and low cost of SCMs and 2) to
environment (often due to de-icing salts or proximity to marine determine how the Pareto-optimal results are determined by this
environments), diffuse into concrete. When a high enough design scenario. For instance, due to high chloride exposure, it is
concentration of chloride ions reaches the concrete expected that the Pareto-optimal solutions may exhibit relatively
reinforcement, the reinforcement corrodes, which produces high SCM replacement percentages or low w/b ratios in order to
expansive rust products and subsequent concrete cracking and generate concrete mixes that resist chloride diffusion for 50 years.
spalling. Specifically, the concrete structure being analyzed is assumed to
Modeling of chloride-induced corrosion has been discussed be near the coastline with exposure to marine spray, which
extensively in the literature (Srubar, 2015; Sohanghpurwala, 2006; involves a maximum chloride boundary condition of 23.5 kg/
Liu and Weyers, 1998; Bentz, 2003; Cusson et al., 2003). In m3 and a 10-years ramp up to this value (see supplementary
previous research, expected service life (ts) with regard to information) (Ehlen et al.). This case study is considered a high
corrosion-induced cracking (t) can be estimated using a two- chloride exposure scenario. In addition, temperatures in Miami
part damage model first proposed by Tuutti in which the time to are relatively high, which facilitates rapid chloride diffusion. In
rebar cracking is the sum of the time to corrosion initiation plus Miami, fly ash is significantly cheaper than slag or cement at a
the time to concrete cracking (Tuutti, 1982). Life-365™ is price of $80 per tonne. Slag is $110 per tonne, while cement is
advanced chloride corrosion modeling software, which $120 per tonne. Transportation of fly ash and slag is highly local
accounts for both time and temperature-dependent changes in as is shown in Table 5, meaning that emissions due to
the rate of chloride diffusion. In addition, it uses the Crank- transportation of SCMs will be low compared to other case
Nicolson finite difference method to solve for t. In this study, we studies.
utilize the same methodology as Life-365™ in the simulation- In the San Francisco case study (Case 2), the purpose is to
optimization framework (Ehlen et al., 2009). For a full discussion investigate a scenario in which the availability of SCMs is low. San
of the service life model equations and parameters used in this Francisco has a high demand but a low local supply of SCMs,
analysis, see the Supplementary Information. which drives up the cost of fly ash and slag to that of cement. In
addition, in this scenario, we mimic the low supply of fly ash and
Case Studies slag by assuming that these materials are imported from
In this work, we apply the simulation-optimization framework international suppliers (Indian Bureau of Mines, 2010; The
described in the previous section to three case studies to illustrate Morning Call, 2017). Thus, the carbon dioxide emissions
how it can be applied to specific concrete design problems. For contributed by fly ash and slag is expected to be higher
comparability, certain aspects of the problem formulation are compared to the Miami case study. In addition, the concrete

Frontiers in Materials | www.frontiersin.org 5 July 2021 | Volume 8 | Article 680895

DeRousseau et al. Multi-Objective Optimization of Low-Carbon Concrete

TABLE 5 | Summary of input parameters specific to each case study.

Input parameter Parameter Case 1 -Miami, FL Case 2-San Cases 3a—Anchorage, AK Case 3b-
group Francisco, CA Anchorage, AK

Costs Cement $115 $120 $180 $180

Fly ash $80 $116 $210 $210
Slag $110 $112 $210 $210
Chloride boundary Exposure scenario Marine spray 1.5 km from shore Marine spray Marine spray
Condition (Ehlen et al.) Maximum concentration 23.5 kg/m3 14.1 kg/m3 23.5 kg/m3 23.5 kg/m3
Time to max 10 years 30 years 10 years 10 years
Transportation Primary transportation method Truck Ship + truck Ship + truck Ship + truck
Fly ash distance 62 km 9,997 km, 20 km 2,642 km, 40 km 2,642 km, 40 km
Slag distance 15 km 8,671 km, 20 km 2,642 km, 40 km 2,642 km, 40 km
Cement distance 124 km 124 km 124 km 124 km
Average monthly January 19.6 °C 9.3 °C −9.5 °C −4.0 °C
Temperature February 20.4 °C 11.2 °C −7.4 °C −2.2 °C
March 22.1 °C 11.8 °C −3.5 °C 0.6 °C
April 24.0 °C 13.1 °C 2.1 °C 5.6 °C
May 25.9 °C 14.5 °C 8.1 °C 11.1 °C
June 27.4 °C 16.4 °C 12.4 °C 13.1 °C
July 28.2 °C 17.1 °C 14.7 °C 16.1 °C
August 28.2 °C 17.6 °C 13.5 °C 16.1 °C
September 27.7 °C 18.1 °C 9.1 °C 11.7 °C
October 25.7 °C 16.1 °C 1.4 °C 5.0 °C
November 23.1 °C 12.7 °C −6.0 °C 0.3 °C
December 20.6 °C 9.7 °C −8.7 °C 2.2 °C
Service life Service life requirement 50 years 50 years 150 years 150 years
Climate change Accounting for climate change? No No No Yes

TABLE 6 | Qualitative summary of input parameters specific to each case study.

Input parameter Case 1-Miami, FL Case 2-San Francisco, CA Case 3—Anchorage Alaska

Relative cost of cementitious materials Low cost of fly ash High cost of both fly ash and slag Very high cost of both fly ash and slag
Chloride exposure scenario Severe chloride exposure Mild chloride exposure Severe chloride exposure
SCM transportation Short distance for SCM transport Far distance for SCM transport Far distance for SCM transport
Average monthly temperature Hot Mild Cold
Service life requirement 50 years 50 years 150 years
Climate change accounting No No Yes

building is assumed to be 1.5 km from the coast and thus receives increased temperatures due to climate change will increase
a mild chloride exposure boundary condition of 14.1 kg/m3 and a chloride diffusion rates compared to the “base case” This
10-years ramp up to this value (Ehlen et al.). San Francisco also specific temperature data is included in Table 5. We
has lower average temperatures than Miami. Consequently, less hypothesize that increases in temperature will increase
rapid chloride diffusion is expected over 50 years. chloride diffusion rates and cause more rapid chloride-induced
Finally, in the Anchorage scenario (Case 3), the goal is to corrosion. In this case study, the concrete structure is assumed to
investigate the impact of accounting for the predicted increase in be on the coast exposed to a marine spray exposure condition
average temperatures related to climate change over a longer (i.e., 23.5 kg/m3, 10-years ramp up period). As seen in Table 5 the
service life. In this case study, we analyze the Pareto optimal high costs and transportation distances for SCMs are due to the
concrete mixture designs for a scenario in which historical lack of local supply.
monthly average temperatures are used (Case 3a) compared to
a scenario that uses predictions for increases in monthly average
temperature using a regional climate prediction model for Alaska RESULTS
(Case 3b) (Alaska Regional Climate Projections, 2009). Note that
temperatures linearly increase from 2020 to the predicted As anticipated, the Pareto-optimal solutions for each case
temperature for 2,100 in the diffusion model and continue to study vary significantly due to the different design
increase linearly throughout the 150-years service life. Since the parameters in each location. In the parallel axis plots,
diffusion model accounts for monthly average temperatures as an values of the objectives and the decision variables are
input to calculate the rate of chloride diffusion, the expected provided to illustrate their connection. All three objectives

Frontiers in Materials | www.frontiersin.org 6 July 2021 | Volume 8 | Article 680895

DeRousseau et al. Multi-Objective Optimization of Low-Carbon Concrete

uses slag, and black lines indicate a solution that uses

no SCMs.
In Figure 3, the four axes on the left represent the decision
variables and the three axes on the right represent the objectives.
The axes for the objectives are oriented such that the preferred
direction is down. Note that there is no preferred direction for the
decisions. In addition, each solution is represented with a line
connecting all the axes. Finally, crossed lines indicate a conflict
between objectives for those two solutions. In other words, a
lower magnitude of one objective (better performance) is related
to a higher magnitude of a second objective (worse performance).
As an example, the crossed lines between, the cost and
compressive strength objectives indicate that for a given
solution, better performance with respect to one of the
FIGURE 1 | Conceptualization of the simulation-optimization framework objectives will mean worse performance in the other.
for analyzing each scenario. The Borg MOEA is coupled with the simulation of
Looking at the objective space, there is a conflict between
concrete properties to find Pareto-optimal mixture designs for different
scenarios.
compressive strength and the other two objectives (i.e., cost,
embodied carbon), as anticipated. High compressive strength
concrete mixtures also exhibit high costs and high embodied
are oriented such that preferred values of each objective are in carbon values; the reverse is also true. Cost and embodied carbon
the downward direction. tend to not conflict; in other words, low-cost solutions tend to
also have low embodied carbon, given that an increase in
Case 1: Miami, Florida Portland cement is the main driver for an increase in both
In the Miami scenario, we seek to determine the set of viable cost and embodied carbon. Other research and industry
mixture designs for a severe chloride exposure condition in a groups have noted these tradeoffs previously (Fantilli et al.,
SCM market with low costs and transportation distances. 2009). In terms of decision variables, the low-cost solutions
Figure 2 illustrates the full set of Pareto-optimal designs tend to be associated with a high w/b ratio, high SCM
solutions for this case study with a 50-years service life replacement percentages, and low total quantity of
constraint. In the figure, red lines indicate a solution in cementitious materials. High strength solutions are associated
which fly ash is used, blue lines indicate a solution that with the opposite. Low embodied carbon mixtures are most

FIGURE 2 | Selection of three possible mix designs to illustrate trade-offs.Set of optimal mixture designs for Miami case study.

Frontiers in Materials | www.frontiersin.org 7 July 2021 | Volume 8 | Article 680895

DeRousseau et al. Multi-Objective Optimization of Low-Carbon Concrete

FIGURE 3 | Set of optimal mixture designs for Miami case study.

greatly affected by the overall quantity of Portland cement, so low note that, due to the inherent tradeoffs between these objectives,
total cementitious content and high SCM replacement that choosing a solution with a ‘desirable’ value for one objective
percentages can provide low embodied carbon. Looking at the means that one (or more) of the other objectives necessarily
decision variable space, it is apparent that the respective limits on becomes less desirable.
fly ash and slag (30 and 50%) are both reached. Recall from
Table 5 that fly ash is 27% cheaper than slag and 33% cheaper Case 2: San Francisco, California
than cement in the Miami scenario; the low cost of fly ash likely The goal of this case study was to elucidate optimal mixture designs
contributes to its usage in the concrete mixtures with the lowest for a case study location that has 1) a mild chloride exposure
cost. Alternatively, note that the mixture with the lowest condition and 2) high costs and transportation distances for
embodied carbon utilizes slag; this result is due to the fact that SCMs. Figure 4 illustrates the set of optimal solutions for this
a higher slag replacement is allowed (up to 50%). Finally, note case study. One key result from this scenario is that none of the
that all mixtures utilize some level of SCM. We hypothesize that optimal mixture designs reach the maximum allowable SCM
this result is due to the existence of severe chloride exposure replacement percentages. Moreover, unlike the Miami case
conditions and inexpensive, local SCMs, which are well known to studies, there are several solutions which utilize less than 10%
densify concrete microstructures and improve the chloride SCM replacement for both fly ash and slag and several that use
resistance of concrete via the pozzolanic reaction. no SCMs whatsoever. These results are likely due to two
In Figure 2 (and subsequently listed in Table 7) we illustrate compounding factors. First, fly ash and slag are nearly the same
via three examples how a concrete designer might use these cost as cement and have high transportation emissions, which
results to select a Pareto-optimal concrete based on preferences reduces their utility for reducing cost or embodied carbon.
for the project. First, suppose that a concrete designer deemed the Second, Case 1 does not incentivize the use of SCMs as much as
embodied carbon objective as the most critical. In this case, Mix A Case 2 due to the mild chloride exposure conditions.
would be selected because it is a mixture that satisfies the problem Furthermore, we see that the objective values of the solutions can be
formulation constraints and has the minimum possible embodied broadly classified into two groups. The first group includes solutions
carbon. Second, suppose that the concrete designer required a that use fly ash, which tend to have relatively high embodied carbon,
50 MPa mixture and that the most important objective was to high strength, and high cost. Looking at the corresponding decision
minimize cost. In this case, Mix B would be optimal because it variables, these solutions use relatively higher quantities of total
satisfies the requirement for compressive strength and among the cementitious materials. In the other group are mixtures that use
remaining feasible solutions, mix B has the lowest cost. Third, lower quantities of cementitious materials and either slag or no
consider a case where the most critical objective is cost. In this SCMs. In the objective space, these variables tend to be low cost
case, Mix C would be selected because it has the absolute lowest and low embodied carbon, but also lower strength than the group of
cost among the set of Pareto-optimal solutions. It is important to solutions that use fly ash. These trends are significantly different than

Frontiers in Materials | www.frontiersin.org 8 July 2021 | Volume 8 | Article 680895

DeRousseau et al. Multi-Objective Optimization of Low-Carbon Concrete

TABLE 7 | Three example concrete mixture designs for the Miami case study.

Mix Cost Compressive Embodied Total W/b c/a SCM Percent

name ($/m3) strength carbon SCMs ratio ratio type replacement
(MPa) (kgCO2/m3) (kg) (%)

Mix A 87.4 45.1 174.0 200 0.53 65.8 Slag 39.3

Mix B 87.2 53.2 332.7 401.1 0.75 74.82 fly ash 17.6
Mix C 80.7 32.7 175.3 200 0.75 55.1 fly ash 30.0

FIGURE 4 | Set of optimal mixture designs for San Francisco case study.

the trends seen in the Miami case study, in which most Pareto-optimal change will cause a more rapid violation of the service life
solutions employed fly ash replacement. We hypothesize that the constraint in Case 3b.
comparably high cost of both fly ash and slag in this case study Figure 5Areports the set of optimal mixture designs for the
incentivized the smaller quantities of SCM replacement. Case 3a. Notably, the total costs of these mixtures are significantly
Another outcome that was unexpected is that the Pareto- higher than that of the Miami and San Francisco scenarios, due to
optimal mixture design with the lowest embodied carbon the high costs of cement, fly ash, and slag in the Anchorage market.
does not utilize SCMs whatsoever in this case study. We Regarding the decision variables, we see that the limits to fly ash
hypothesize that this result is due to the fact that this mixture and slag replacement are not reached—a result attributable to the
design uses the lowest allowable quantity of total fact that slag and fly ash are 10% more expensive than cement. Also
cementitious materials and because (compared to the other note the existence of several concrete mixture designs that use only
case studies) fly ash and slag are actually associated with high cement—again, an interesting, but expected, finding in this market
embodied carbon emissions due to the assumed where the cost of cement is less than either fly ash or slag.
transportation distances. It appears that the more severe service life condition in Case 3b
(due to temperature rise) has caused a change in the set of optimal
Case 3: Anchorage, Alaska mixture designs. Figure 5B illustrates the set of optimal mixture
The purpose of the Anchorage case study was to investigate designs when higher average temperatures are considered.
whether accounting for increased average temperatures due to Looking at the objective space, we see that the MOEA was
climate change alters the set of Pareto-optimal mixture designs. unable to find a 35 MPa mixture that was feasible and non-
In Case 3a we use historical monthly average temperatures to dominated because this solution uses a high w/b ratio and low
quantify chloride ingress. In Case 3b, we employ predicted quantity of total cementitious materials. In addition, several
average temperatures for Anchorage using the regional climate solutions which used only cement have become infeasible due
model discussed in Case Studies. Since increased temperatures to the service life constraint. Despite the fact that cement is the
cause increased rates of diffusion, it was expected that climate least expensive cementitious material in Case 3b, the amount of

Frontiers in Materials | www.frontiersin.org 9 July 2021 | Volume 8 | Article 680895

DeRousseau et al. Multi-Objective Optimization of Low-Carbon Concrete

FIGURE 5 | Pareto-optimal solutions for (A) Case 3a and (B) Case 3b, which illustrates the importance of climate change accounting regarding prevention of
corrosion in chloride-laden conditions.

SCM use tends to increase due to the need to prevent accelerated designs with compressive strengths around the 55 MPa
corrosion that can arise from increased temperatures. utilize fly ash instead of slag (or only cement) in order to
Furthermore, it was found that an increase in the required meet the increased durability requirement. We see that these
durability of a concrete appears to shift the balance between fly mixtures use high cementitious materials quantities, which
ash and slag utility for this case study. Concrete mixture leads to high values of embodied carbon, and moderately high

Frontiers in Materials | www.frontiersin.org 10 July 2021 | Volume 8 | Article 680895

DeRousseau et al. Multi-Objective Optimization of Low-Carbon Concrete

cost. These results point to the sensitivity of the optimal modeling these mechanisms and how they are affected by the
solutions for a given case study to increased temperatures. quantity and types of mixture constituents used. Simulating
However, it should be noted that this sensitivity will be multiple degradation mechanisms will also provide concrete
different for every design scenario. designers with information about which mechanism is expected
to cause degradation first. For example, if a concrete is placed in an
environment in which both freeze-thaw conditions and chloride-
DISCUSSION exposure conditions occur, modeling of both degradations can help
determine which mechanism is the most critical.
General Trends Second, future work should also consider how other types of
Due to their advantages for strengthening and densifying concrete, SCMs besides fly ash and slag (e.g., silica fume, metakaolin) impact
SCMs (especially fly ash) are commonly used in scenarios where simulated concrete properties. Novel supplementary cementitious
prevention of chloride-induced corrosion is necessary. For materials have been the subject of increased research (Juenger and
instance, in the two scenarios with extreme chloride exposure Maria, 2019) given global decarbonization and decommissioning of
conditions (Miami and Anchorage) Miami case study, there is a coal-fired power plants that produce large quantities of beneficial fly
prevalence of optimal mixture designs that employ SCMs. ash at a time in which the demand for SCMs continues to increase.
Furthermore, comparing Case 3a to Case 3b, we see that the However, to integrate alternative SCMs into this simulation-
increased durability concerns caused several concrete mixtures optimization framework, first the impact of the type and quantity
without SCMs to become infeasible, despite the cost premium for of each SCM on cost, compressive strength, embodied carbon, and
SCMs compared to cement in the Anchorage market. service life must be modeled. These additional models will need to be
Furthermore, we see a trend in the decision variables that are developed using machine learning, life cycle assessment, and
used to design high-strength mixtures. In all cases, fly ash is the physics-based service life modeling. Similarly, incorporating
SCM selected for high strength mixtures; specifically, all cases use models that assess the impacts of adding multiple SCMs to
fly ash to design mixtures above 55 MPa. High strength concrete concrete mixtures (i.e., ternary blends) would be beneficial for
mixtures also used high quantities (>340 kg) of cementitious deciding what SCMs to use for different types of mixture design
materials in all these cases. Relatively low w/b ratios were also priorities (Khan, 2012). The use of alternative SCMs would have
seen to be an important factor for designing high strength similarly beneficial impacts on compressive strength and service life
concrete (0.30–0.55). do to their pozzolanic nature (Behnood and Ziari, 2008).
Another general trend is that the mixtures with the lowest Thrd, as a greater variety of SCMs and degradation mechanisms
embodied carbon contained slag, which was observed in all case are incorporated into the simulation-optimization framework,
study examples. This phenomenon is likely because slag more data will be needed for modeling these relationships. For
replacement was defined to be as high as 50%, compared to instance, the development of machine learning models for
30% for fly ash. This low-carbon trend was consistent, regardless predicting compressive strength requires broad sampling of the
of whether slag was more expensive or had higher transportation predictor variable space to make accurate predictions. Future
emissions than fly ash. Of course, this result would change if the research should aim to develop publicly available databases that
allowable replacement percentages were increased for fly ash. contain mixture design information such as the quantities and
Low-cost concrete mixtures tended to use low quantities of total properties of the mixture constituents, as well as resultant
cementitious materials; however, these mixtures were not properties of the mixture such as compressive strength and
consistent for all scenarios in terms of fly ash or slag use slump. As an example of this kind of database, the University
because of the changing relative cost of cementitious materials. of California Irvine’s Machine Learning Repository houses a
It was found that both embodied carbon and cost conflict with the dataset of 1,000 concrete mixture designs and the resultant
compressive strength objective in all scenarios. compressive strength (UCI Machine Learning Repository, 2020).
This dataset should be expanded to include mixture design
Future Opportunities information including other types of SCMs and constituent
In the section that follows, we outline four ways that this research material properties.
could be advanced. First, the service life modeling should be Last, it would be beneficial for the optimization analyses to
expanded. Recall that in this study, service life is predicted for directly consider uncertainty for each constraint and objective.
concretes in chloride-laden environments. Although chloride- Currently, the compressive strength and embodied carbon
induced corrosion is the costliest degradation mechanism in the models quantify the average error, but the only way for
U.S., there are other types of concrete degradation mechanisms, concrete designers to use this uncertainty information is to
which may be dominant under certain conditions. For instance, build safety factors into final mixture designs. A direct method
when concrete is exposed to water containing sulfate ions (e.g., from for considering uncertainty could involve a technique called
drainage, sewers), sulfate ions diffuse into concrete, react with many objective robust decision making in which multi-
cement hydration products, and cause cracking. Other objective optimization is used to determine a set of Pareto-
degradation mechanisms can include freeze-thaw damage, alkali- optimal solutions; then these solutions are evaluated under a
silica reaction, alkali-carbonate reaction, delayed-ettringite variety of scenarios to determine which solutions perform well
formation, among many others. Equations should be (i.e., are robust) under many circumstances (Kasprzyk and Nataraj,
incorporated in the simulation-optimization framework for 2013; Kasprzyk and Reed, 2012; Kasprzyk, Nataraj, et al.).

Frontiers in Materials | www.frontiersin.org 11 July 2021 | Volume 8 | Article 680895

DeRousseau et al. Multi-Objective Optimization of Low-Carbon Concrete

CONCLUSION DATA AVAILABILITY STATEMENT

This study presents the formulation and implementation of The raw data supporting the conclusions of this article will be
a simulation-optimization framework in Python that integrates made available by the authors, without undue reservation.
predictive property models to design optimal concrete
mixtures under many design scenarios. The simulation-
optimization framework developed primarily improves upon AUTHOR CONTRIBUTIONS
the existing literature because it enables designers to find
optimal concrete mixture designs while simultaneously MD, JK, and WS contributed conceived and designed study. MD
considering environmental, economic, durability, and conducted the study and wrote the first draft of the article. All
hardened-state performance. authors contributed to article editing. All authors approved the
The utility of the simulation-optimization framework submitted version.
was tested for several concrete design case studies (based
in Miami, San Francisco, and Anchorage) and the objectives
of compressive strength, embodied carbon, and cost were FUNDING
optimized. Results from these scenarios demonstrate that
the local conditions of the case study dictate the most This research was made possible the Department of Civil,
important parameters of the simulation-optimization Environmental, and Architectural Engineering, the College of
(i.e., relative constituent costs, in situ service-life Engineering and Applied Sciences, and the Living Materials
conditions). Out of all other input parameters, constituent Laboratory (LMLab) at the University of Colorado Boulder,
cost and service-life conditions most greatly impact the with support from the National Science Foundation (Award
set of optimal concrete mixture designs in terms of the No. CMMI-1562557). This work represents the views of the
types and quantities of mixture ingredients that are authors and not necessarily those of the sponsors.
utilized. For instance, high costs for SCMs produced
Pareto-optimal mixture designs with significantly lower
SCM replacement percentages. Similarly, more aggressive ACKNOWLEDGMENTS
chloride exposure conditions caused optimal concrete
mixtures to have lower w/b ratios and higher SCM Publication of this article was funded by the University of
replacement percentages to densify the concrete and Colorado Boulder Libraries Open Access Fund.
meet the service life requirement. The highly varying
solution sets for the case studies illustrate that the
simulation-optimization framework can be utilized to SUPPLEMENTARY MATERIAL
quantify and visualize tradeoffs between critical concrete
performance metrics. Such a tool can be used to tailor The Supplementary Material for this article can be found online at:
low-carbon concrete mixtures to the exact preferences of https://www.frontiersin.org/articles/10.3389/fmats.2021.680895/
the designer. full#supplementary-material

Coello, C. A., Lamont, G. B., and Van Veldhuizen, D. A. (2007). Evolutionary

REFERENCES Algorithms for Solving Multi-Objective Problems. New York, NY: Springer, Vol. 5.
Cusson, D., Lounis, Z., and Daigle, L. (2011). Durability Monitoring for Improved
American Concrete Institute (1991). ACI 211.1 - Standard Practice for Selecting Service Life Predictions of Concrete Bridge Decks in Corrosive Environments.
Proportions for Normal, Heavyweight, and Mass Concrete. Farmington Hills, Computer-Aided Civil Infrastructure Eng. 26 (7), 524–541. doi:10.1111/j.1467-
MI: American Concrete Institute. 8667.2010.00710.x
Alabama Standard Specifications (2012). 2012 Edition - Portland Cement Concrete DeRousseau, M. A., Arehart, J. H., Kasprzyk, J. R., and Srubar, W. V. (2020).
Pavement, Section 450. Alabama Department of Transportation. Available at: Statistical Variation in the Embodied Carbon of Concrete Mixtures. J. Clean.
https://www.dot.state.al.us/conweb/pdf/Specifications/2012%20GASP%20Summary/ Prod. 275, 123088. doi:10.1016/j.jclepro.2020.123088
12-0152.pdf (Accessed May 12, 2020). DeRousseau, M. A., Kasprzyk, J. R., and Srubar, W. V. (2018). Computational
Alaska Regional Climate Projections (2009). Alaska Regional Climate Projections. Design Optimization of Concrete Mixtures: A Review. Cement Concrete Res.
Fairbanks, AK: University of Alaska Fairbanks. Available at: https://www.uaf. 109, 42–53. doi:10.1016/j.cemconres.2018.04.007
edu/files/snre/MP_09_04.pdf. DeRousseau, M. A., Laftchiev, E., Kasprzyk, J. R., Rajagopalan, B., and Srubar III,
Baykasoğlu, A., and Özbay, E. (2009). Prediction and Multi-Objective W. V. (2019). A Comparison of Machine Learning Methods for Predicting the
Optimization of High-Strength Concrete Parameters via Soft Computing Compressive Strength of Field-Placed Concrete. Construction Building Mater.
Approaches. Expert Syst. Appl. 36 (3), 6145–6155. doi:10.1016/ 228, 116661. doi:10.1016/j.conbuildmat.2019.08.042
j.eswa.2008.07.017 Domone, P. (1998). The Slump Flow Test for High-Workability Concrete
Behnood, A., and Ziari, H. (2008). Effects of Silica Fume Addition and Water to 11Communicated by C.D. Pomeroy. Cement Concrete Res. 28 (2), 177–182.
Cement Ratio on the Properties of High-Strength Concrete after Exposure to doi:10.1016/S0008-8846(97)00224-X
High Temperatures. Cement Concrete Composites 30 (2), 106–112. doi:10.1016/ Ehlen, M. A., Michael, D. A. T., and Evan, C. B. (2009). Life-365 Service Life
j.cemconcomp.2007.06.003 Prediction ModelTM Version 2.0. Concrete Int. 31 (5), 41–46.
Bentz, E. C. (2003). Probabilistic Modeling of Service Life for Structures Subjected Fantilli, A. P., Mancinelli, O., and Chiaia, B. (2019). The Carbon Footprint of
to Chlorides. Mater. J. 100 (5), 391–397. doi:10.14359/12814 Normal and High-Strength Concrete Used in Low-Rise and High-Rise

Frontiers in Materials | www.frontiersin.org 12 July 2021 | Volume 8 | Article 680895

DeRousseau et al. Multi-Objective Optimization of Low-Carbon Concrete

Buildings. Case Stud. Construction Mater. 11, e00296. doi:10.1016/ D.C.: National Cooperative Highway Research Program. Available at: https://
j.cscm.2019.e00296 www.nap.edu/read/13934/chapter/1.
Jazmin, Z. S., Patrick, M. R., Jonathan, D. H., Matteo, G., and Andrea, C. (2016). A Srubar, W. V. (2015). Stochastic Service-Life Modeling of Chloride-Induced
Diagnostic Assessment of Evolutionary Algorithms for Multi-Objective Surface Corrosion in Recycled-Aggregate Concrete. Cement Concrete Composites 55
Water Reservoir Control. Adv. Water Resour. 92, 172–185. doi:10.1016/ (Jan), 103–111. doi:10.1016/j.cemconcomp.2014.09.003
j.advwatres.2016.04.006 The Morning Call (2017). Despite Glut of Coal Ash, U.S. Is Importing it from
Juenger, M. C. G., Snellings, R., and Bernal, S. A. (2019). Supplementary Other Countries. Associated Press. Availabe at: https://www.mcall.com/
Cementitious Materials: New Sources, Characterization, and Performance business/mc-despite-glut-of-coal-ash-u-s-is-importing-it-from-other-countries-
Insights. Cement Concrete Res. 122, 257–273. doi:10.1016/ 20170323-story.html (Accessed June 4, 2020).
j.cemconres.2019.05.008 Toutanji, H., Delatte, N., Aggoun, S., Duval, R., and Danson, A. (2004). Effect of
Kasprzyk, J. R., Nataraj, S., Reed, P. M., and Lempert, R. J. (2013). Many Objective Supplementary Cementitious Materials on the Compressive Strength and
Robust Decision Making for Complex Environmental Systems Undergoing Durability of Short-Term Cured Concrete. Cement Concrete Res. 34 (2),
Change. Environ. Model. Softw. 42, 55–71. doi:10.1016/j.envsoft.2012.12.007 311–319. doi:10.1016/j.cemconres.2003.08.017
Kasprzyk, J. R., Reed, P. M., Characklis, G. W., and Kirsch, B. R. (2012). Many- Tuutti, K. (1982). Corrosion of Steel in Concrete. Lund, Sweden: Cement-och
objective De Novo Water Supply Portfolio Planning under Deep Uncertainty. betonginst.
Environ. Model. Softw. 34, 87–104. doi:10.1016/j.envsoft.2011.04.003 UCI Machine Learning Repository (2020). UCI Machine Learning Repository:
Khan, M. I. (2012). Predicting Properties of High Performance Concrete Concrete Compressive Strength Data Set. Irvine, CA. Available at: https://
Containing Composite Cementitious Materials Using Artificial Neural archive.ics.uci.edu/ml/datasets/Concrete+Compressive+Strength (Accessed
Networks. Automat. Construct. 22, 516–524. doi:10.1016/j.autcon.2011.11.011 Oct 31, 2018).
Khatri, R. P., Sirivivatnanon, V., and Gross, W. (1995). Effect of Different Wilson, M. L., and Kosmatka, S. H. (2011). Design and Control of Concrete
Supplementary Cementitious Materials on Mechanical Properties of High Mixtures. 15 edition. Skokie, IL: Portland Cement Assn.
Performance Concrete. Cement Concrete Res. 25 (1), 209–220. doi:10.1016/ Young, B. A., Hall, A., Pilon, L., Gupta, P., and Sant, G. (2019). Can the
0008-8846(94)00128-L Compressive Strength of Concrete Be Estimated from Knowledge of the
Liu, T., and Weyers, R. W. (1998). Modeling the Dynamic Corrosion Process in Mixture Proportions?: New Insights from Statistical Analysis and Machine
Chloride Contaminated Concrete Structures. Cement and Concrete Research 28 Learning Methods. Cement Concrete Res. 115, 379–388. doi:10.1016/
(3), 365–379. doi:10.1016/S0008-8846(98)00259-2 j.cemconres.2018.09.006
Madurwar, M., Sakhare, V., and Ralegaonkar, R. (2015). Multi Objective
Optimization of Mix Proportion for a Sustainable Construction Material. Conflict of Interest: The authors declare that the research was conducted in the
Proced. Eng. 118, 276–283. doi:10.1016/j.proeng.2015.08.427 absence of any commercial or financial relationships that could be construed as a
Indian Bureau of Mines (2010). Minerals Yearbook. Nagpur, India: Bureau of potential conflict of interest.
Mines.
Promentilla, M. A. B., Kalaw, M. E., Nguyen, H. T., Aviso, K. B., and Tan, R. R. Publisher’s Note: All claims expressed in this article are solely those of the authors
(2017). A Fuzzy Programming Approach to Multi-Objective Optimization for and do not necessarily represent those of their affiliated organizations, or those of
Geopolymer Product Design. Computer Aided Chem. Eng. 40, 1015–1020. the publisher, the editors and the reviewers. Any product that may be evaluated in
doi:10.1016/b978-0-444-63965-3.50171-9 this article, or claim that may be made by its manufacturer, is not guaranteed or
Reed, P. M., Hadka, D., Herman, J. D., Kasprzyk, J. R., and Kollat, J. B. (2013). endorsed by the publisher.
Evolutionary Multiobjective Optimization in Water Resources: The Past,
Present, and Future. Adv. Water Resour. 51, 438–456. doi:10.1016/ Copyright © 2021 DeRousseau, Kasprzyk and Srubar. This is an open-access article
j.advwatres.2012.01.005 distributed under the terms of the Creative Commons Attribution License (CC BY).
Rodgers, L. (2018). The Massive CO2 Emitter You May Not Know about. BBC The use, distribution or reproduction in other forums is permitted, provided the
News. Available at: https://www.bbc.com/news/science-environment- original author(s) and the copyright owner(s) are credited and that the original
46455844 (Accessed April 18, 2020). publication in this journal is cited, in accordance with accepted academic practice.
Sohanghpurwala, A. A. (2006). Manual on Service Life of Corrosion-Damaged No use, distribution or reproduction is permitted which does not comply with
Reinforced Concrete Bridge Superstructure Elements. Report 558. Washington, these terms.

Frontiers in Materials | www.frontiersin.org 13 July 2021 | Volume 8 | Article 680895